The Overexpression of Hypomethylated miR-663 Induces Chemotherapy Resistance in Human Breast Cancer Cells by Targeting Heparin Sulfate Proteoglycan 2 (HSPG2)*

Background: miR-663 is related to chemosensitivity in human breast cancer cells. Results: Overexpression of miR-663 was associated with chemoresistance and accompanied by down-regulation of HSPG2. Conclusion: Overexpression of hypomethylated miR-663 induces chemoresistance in breast cancer cells by down-regulating HSPG2. Significance: Learning how miR-663 regulates chemoresistance may provide a potential target for the miRNA-based approach of breast cancer therapy. MicroRNAs are involved in regulating the biology of cancer cells, but their involvement in chemoresistance is not fully understood. We found that miR-663 was up-regulated in our induced multidrug-resistant MDA-MB-231/ADM cell line and that this up-regulation was closely related to chemosensitivity. In the present study, we aimed to clarify the role of miR-663 in regulating the chemoresistance of breast cancer. MicroRNA microarray and quantitative RT-PCR assays were used to identify differentially expressed microRNAs. Cell apoptosis was evaluated by annexin V/propidium iodide staining, TUNEL, and reactive oxygen species generation analysis. The expression of miR-663 and HSPG2 in breast cancer tissues was detected by in situ hybridization and immunohistochemistry. The potential targets of miR-663 were defined by a luciferase reporter assay. Bisulfite sequencing PCR was used to analyze the methylation status. We found that miR-663 was significantly elevated in MDA-MB-231/ADM cells, and the down-regulation of miR-663 sensitized MDA-MB-231/ADM cells to both cyclophosphamide and docetaxel. The overexpression of miR-663 in breast tumor tissues was associated with chemoresistance; in MDA-MB-231 cells, this chemoresistance was accompanied by the down-regulation of HSPG2, which was identified as a target of miR-663. MDA-MB-231/ADM contained fewer methylated CpG sites than its parental cell line, and miR-663 expression in MDA-MB-231 cells was reactivated by 5-aza-29-deoxycytidine treatment, indicating that DNA methylation may play a functional role in the expression of miR-663. Our findings suggest that the overexpression of hypomethylated miR-663 induced chemoresistance in breast cancer cells by down-regulating HSPG2, thus providing a potential target for the development of an microRNA-based approach for breast cancer therapy.


MicroRNAs are involved in regulating the biology of cancer cells, but their involvement in chemoresistance is not fully understood. We found that miR-663 was up-regulated in our induced multidrug-resistant MDA-MB-231/ADM cell line and that this up-regulation was closely related to chemosensitivity.
In the present study, we aimed to clarify the role of miR-663 in regulating the chemoresistance of breast cancer. MicroRNA microarray and quantitative RT-PCR assays were used to identify differentially expressed microRNAs. Cell apoptosis was evaluated by annexin V/propidium iodide staining, TUNEL, and reactive oxygen species generation analysis. The expression of miR-663 and HSPG2 in breast cancer tissues was detected by in situ hybridization and immunohistochemistry. The potential targets of miR-663 were defined by a luciferase reporter assay. Bisulfite sequencing PCR was used to analyze the methylation status. We found that miR-663 was significantly elevated in MDA-MB-231/ADM cells, and the down-regulation of miR-663 sensitized MDA-MB-231/ADM cells to both cyclophosphamide and docetaxel. The overexpression of miR-663 in breast tumor tissues was associated with chemoresistance; in MDA-MB-231 cells, this chemoresistance was accompanied by the down-regulation of HSPG2, which was identified as a target of miR-663. MDA-MB-231/ADM contained fewer methylated CpG sites than its parental cell line, and miR-663 expression in MDA-MB-231 cells was reactivated by 5-aza-29-deoxycytidine treatment, indicating that DNA methylation may play a functional role in the expression of miR-663. Our findings suggest that the overexpression of hypomethylated miR-663 induced chemoresistance in breast cancer cells by down-regulating HSPG2, thus providing a potential target for the development of an microRNA-based approach for breast cancer therapy.
Resistance to chemotherapy remains a major obstacle to effective anti-cancer treatments, resulting in relapse and progression in most malignant tumors. Chemoresistance may be caused by drug efflux by transporters, inactivation by detoxification enzymes, the altered expression of pro-apoptosis proteins, changes in tumor suppressor gene expression, or the increased activity of DNA repair mechanisms (1)(2)(3). Recent evidence has demonstrated that microRNAs (miRNAs) 5 take part in these processes (4,5).
miRNAs are small endogenous noncoding RNAs, consisting of 21-24 nucleotides, that are processed from hairpin structures. These small non-coding single-stranded RNAs regulate gene expression at the post-transcriptional level through imperfect base pairing with the 3Ј-UTR, 5Ј-UTR, or coding regions of the target messenger RNAs (6,7). They regulate the biology of cancer cells, including their chemosensitivity (8,9). Aberrant miRNA expression is involved in a number of molecular pathways related to the mechanisms of chemoresistance. For example, miR-214 regulates ovarian cancer stem cell properties by targeting the p53/Nanog axis (5), and down-regulation of miR-15b or miR-16 results in BCL2 overexpression and multidrug resistance in gastric and breast cancer cells (10). Additionally, a decrease in miR-128 expression leading to Bmi-1 and ABCC5 overexpression is a stem cell-like feature of breast tumor-initiating cells that contributes to chemoresistance in breast cancers (11). Other miRNAs, like miR-663, were recently reported to be down-regulated and were identified to act as tumor suppressors in gastric cancer or as oncogenes in nasopharyngeal carcinoma by directly targeting p21 (12).
In this study, a multiple-drug-resistant cell line, MDA-MB-231/Adriamycin (ADM), was derived from MDA-MB-231 cells by exposing them to gradually increasing concentrations of ADM. We found that miR-663 was up-regulated in MDA-MB-231/ADM and that this up-regulation was correlated with chemoresistance; miR-663 expression was also significantly higher in chemoresistant breast cancer tissue than in chemosensitive samples. We therefore investigated the targets and regulatory mechanisms of miR-663. We predicted and confirmed that heparin sulfate proteoglycan 2 (HSPG2) was targeted by miR-663. Furthermore, we found a difference in DNA methylation between MDA-MB-231/ADM and MDA-MB-231 cells. We suggest that hypomethylated miR-663 overexpression induces chemoresistance in human breast cancer cells by targeting HSPG2.
Differentially Expressed miRNAs Detected by MicroRNA Microarray-MDA-MB-231 and MDA-MB-231/ADM cells were harvested and subsequently analyzed by miRNA microarray. Briefly, miRNA was separated from 30 -50 mg of total RNA isolated using TRIzol reagent (Invitrogen), labeled with the miRCURY TM Hy3 TM /Hy5 TM Power Labeling kit (Exiqon), and hybridized to the miRCURY TM LNA Array (Exiqon, version 11.0). Scanning was performed with the Axon GenePix 4000B microarray scanner. GenePix pro version 6.0 was used to read the raw image intensity. Unsupervised hierarchical clustering was performed on the miRNA expression profile. Each miRNA present in the database was mapped to a precise location in the human genome using a BLAST search with the default parameters and the maps available from the National Center for Biotechnology Information Human Genome Resources. In addition, to validate our data, we identified the clones corresponding to each miRNA and mapped them to the human genome.
Quantitative Real-time PCR-Total RNA from treated cells was extracted using TRIzol reagent (Invitrogen) and quantified with an ultraviolet spectrophotometer (UVP Inc.) at a wavelength of 260 nm. The miRNA real-time PCR quantification process included two steps: reverse transcription (RT) and realtime PCR. The primers for the mature miR-663 miRNA and U6 snRNA were obtained from Ambion (Austin, TX). First, the stem-loop RT primer was hybridized to an miRNA molecule, followed by reverse transcription with Moloney murine leukemia virus reverse transcriptase. Next, the RT products were quantified using conventional TaqMan PCR (Bio-Rad). The 20-l RT reaction was incubated at 25°C for 30 min, followed by 94°C for 3 min. The reaction then underwent 45 cycles of denaturation at 94°C for 20 s, annealing at 50°C for 25 s, and synthesis at 72°C for 20 s, as recommended by the manufacturer.
Western Blot-Protein extracts were resolved by 8 -12% SDS-PAGE, transferred to PVDF membranes, and probed with rabbit monoclonal antibody against HSPG2 (Abcam). Peroxidase-conjugated anti-rabbit IgG (Cell Signaling Technology) was used as the secondary antibody, and the antigen-antibody reaction was visualized by the enhanced chemiluminescence assay (ECL).
Measurement of Cell Viability by Cholecystokinin Octapeptide (CCK8) Assay-All cells were seeded in 96-well plates at a density of 2 ϫ 10 4 cells/well. After transfection with an miR-663 mimic or ASO for 12 h, docetaxel (DTX) or cyclophosphamide (CTX) was added, and after growth for 24, 48, and 72 h, 10 l of the CCK8 reagent were added to each well. The plates were incubated for 4 h at 37°C, and the difference in absorbance between 450 and 590 nm was measured as an indicator of cell viability.
Reactive Oxygen Species (ROS) Generation Analysis by Flow Cytometry-The cells were plated in 6-well plates at a density of 2 ϫ 10 6 cells/well. The cells were collected and incubated with a 10 M concentration of the fluorescent probe 2Ј,7Ј-dichlorodihydrofluorescein diacetate for 30 min at 37°C in the dark to measure the accumulation of intracellular ROS. After incubation, the cells were washed with PBS and analyzed within 30 min using a FACScan (BD Biosciences) equipped with an aircooled argon laser tuned to 488 nm. The specific fluorescence signals corresponding to 2Ј,7Ј-dichlorodihydrofluorescein diacetate were collected with a 525-nm band pass filter. A total of 10,000 cells were counted in each assay.
Apoptosis Analysis by Annexin V/Propidium Iodide (PI) Staining-Apoptotic and dead cell counts were performed using FITC-labeled annexin V and PI staining (BD Biosciences), followed by flow cytometry. The cells were gently vortexed and resuspended in binding buffer at a concentration of 3 ϫ 10 6 /ml, and then 100 l of cell suspension was added to 5 l of annexin V-FITC and 10 l of PI and mixed for 15 min in the dark at room temperature. Next, 400 l of PBS was added to the solution. A FACScan instrument (BD Biosciences) was used to count the cells (1 ϫ 10 3 ) at an excitation wavelength of 490 nm. CellQuest software was used for data collection and processing.
TUNEL Analysis by Fluorescence Microscopy-Apoptotic cells were also visualized by the terminal deoxytransferase-mediated dUTP-biotin nick end labeling (TUNEL) method. This assay was performed using the In Situ Cell Death Detection kit (Roche Applied Science) according to the manufacturer's recommendations. Briefly, cells were fixed with 4% paraformaldehyde for 15 min at room temperature, permeabilized with 25 g/ml proteinase K for 30 min at 37°C, and then incubated with 90 l of labeling solution plus 10 l of enzyme solution at 37°C for 2 h. The cells were then washed three times with PBST for 5 min each and incubated with peroxidase at 37°C for 30 min. Cell images were acquired by fluorescence microscopy (Axio-skop, Zeiss). The number of TUNEL-positive cells per

cells was counted, and the percentage of apoptotic cells was calculated.
Human Apoptosis Antibody Array-We used the RayBio Human Apoptosis Antibody Array kit (RayBiotech) to simultaneously detect the relative levels of 43 apoptosis-related proteins in lysates of MDA-MB-231 cells with or without miR-663 that were treated with 0.25 M ADM. The protein extracts were prepared, and 300 g of each were analyzed according to the kit manufacturer's suggestions.
In Vitro Luciferase Target Report Assay-We constructed PmiR-Check2 plasmids for the miR-663 target HSPG2 3Ј-UTR (NM_005529) containing the wild type (WT) pmiR-HSPG2 3Ј-UTR or a mutant pmiR-HSPG2 in which 3 nucleotides within the "seed sequence" were altered. The primers used to generate the pmiR-HSPG2 (968 bp) are as follows: forward primer, 5Ј-CCGCTCGAGGATGCCACGCTTTGCTGC-3Ј (XhoI); reverse primer, 5Ј-ATAAGAATGCGGCCGCGGCG-GGGATAGCACCGTTTA-3Ј (NotI). The oligonucleotides were annealed and inserted into the pmiR-Check2 vector (Promega). Site-directed mutagenesis was performed using the QuikChange kit (Stratagene). In the mutant construct, the predicted miR-663 binding site, CCCCGCC, was replaced with CACAGCA. The empty vector (pmiR-Check2) was used as a negative control. HEK 293T cells were transfected with 0.2 g of the reporter plasmid and 0.01 g of the pmiR-Check2 control plasmid or a 200 nM concentration of either miR-663 or the NC control (GenePharma Co.) per well in 96-well plates. Following 24 h of incubation, the cells were subjected to a luciferase reporter assay using a luciferase assay system (Promega).
Patients and Tumor Specimens-Thirty-nine patients with primary breast cancer who underwent consecutive neoadjuvant chemotherapy at the Breast Tumor Center of Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, from October 2003 to July 2010 were enrolled in the present study. The tumor specimens were obtained by core-needle biopsy prior to starting therapy. Informed consent was obtained from all patients following a protocol approved by the Ethics Committee of the Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University.
The chemotherapy regimen consisted of four 21-day cycles of AC (doxorubicin, 60 mg/m 2 on day 1; CTX, 600 mg/m 2 on day 1) followed by four 21-day cycles of DTX (100 mg/m 2 on day 1) or six 21-day cycles of TAC (DTX, 75 mg/m 2 on day 1; doxorubicin, 50 mg/m 2 on day 1; CTX, 500 mg/m 2 on day 1). Tumor size was evaluated with ultrasound or computerized tomography scans at base line, 14 -21 days after the last cycle of AC, and 21 days after the last cycle of DTX treatment. The effect of chemotherapy on the tumor was assessed as the threedimensional volume reduction rate or tumor response rate. Tumor response was evaluated by the Response Evaluation Criteria in Solid Tumors (RECIST), which is defined as follows: complete response (CR; disappearance of the disease), partial response (PR; Ն30% reduction), stable disease (SD; Ͻ30% reduction or Յ20% enlargement), or progressive disease (PD; Ն20% enlargement).
In Situ Hybridization-MiR-663 expression was examined in 39 formalin-fixed paraffin-embedded breast cancer tissues by in situ hybridization. Briefly, the tissue slides were prehybridized in a hybridization solution (Boster Co., Wuhan, China) at 59°C for 2 h. Ten picomoles of 5Ј-digoxigenin-labeled miRCURY TM locked nucleic acid detection probes (Exiqon, Denmark) complementary to miR-663 or scrambled microRNA were added and hybridized for 16 h at a temperature of 21°C below the calculated melting temperature of the locked nucleic acid probe. After stringent washing, a mouse monoclonal antidigoxigenin antibody (Abcam) and alkaline phosphatasestreptavidin (Zhongshan Golden Bridge Biotechnology Co., Beijing, China) were used to detect the biotinylated probes. The slides were counterstained with nuclear fast red (Enzo Diagnostics) to visualize the nuclei and then mounted in aqueous mounting medium (Maixin Biotechnology Co., Fuzhou, China). Then tissue sections were scored manually semiquantitatively for cytoplasmic staining. The dominant staining intensity in tumor cells was scored as follows: 0, negative; 1, weak; 2, intermediate; 3, strong.
Immunohistochemistry-Immunohistochemical analysis was performed as reported (13). The paraffin sections were incubated with primary antibody against HSPG2 (1:150; Abcam), For negative control, isotype-matched antibody was applied. Immunohistochemical scoring was performed without prior knowledge of the clinical response. Tissue sections were observed under a ZEISS AX10-Imager A1, and all pictures were captured using AxioVision version 4.7 microscopy software. DNA Methylation Analysis-CpG islands were predicted using the UCSC Genome Browser. The miR-663 promoter is located in a region with significant CpG dinucleotide enrich-ment. Genomic DNA from MDA-MB-231 and MDA-MB-231/ ADM cells was prepared using the MethylDetector kit (Active Motif Co.). The PCR primers were designed as follows with Methyl Primer Express version 1.0 software: forward primer, 5Ј-ATAACCAGCCACCTTGATCTG; reverse primer, 5Ј- ATTCGCCAATGACAAGACG. The PCR products were cloned using a T4 ligase cloning kit (Takara), and single clones were sent for sequencing. miR-663 expression in MDA-MB-231 cells with or without 3 mM 5-aza-2Ј-deoxycytidine (5-Aza-dC; Sigma) treatment for 72 h was analyzed by qRT-PCR.
Statistical Analysis-The data are expressed as the mean Ϯ S.D. of at least three independent experiments. Student's t tests and one-way analysis of variance were used to test the differences between the means, whereas the 2 test was applied to compare percentages. All p values are two-sided, and a value of 0.05 was considered to be statistically significant. All statistical calculations were performed using SPSS software (version 13.0; SPSS).

RESULTS
The Establishment of the ADM-resistant Cell Line MDA-MB-231-Over a period of 6 months, MDA-MB-231 cells in culture were continuously exposed to increasing concentrations of ADM. As shown in Fig Table 1). The greatest up-regulation was observed for miR-663, which exhibited a 7.26-fold increase in expression in MDA-MB-231/ADM compared with the parental line. We then measured the miR-663 levels in the different generations of MDA-MB-231/ADM using qRT-PCR. miR-663 expression was dramatically increased, more than 5-fold, in the seventh generation of MDA-MB-231/ADM compared with MDA-MB-231, implying that miR-663 increased sharply under stringent drug stress and was related to drug sensitivity (Fig. 1B). In addition, we also measured the miR-663 levels in other breast tumor cell lines, including BT-474, MDA-MB-436, SKBR3, BT-549, and MCF-7. Because the level of miR-663 in MCF-7 was almost 5-fold higher than that in BT-474, we chose this pair of cell lines, MCF-7 and BT-474, to confirm our hypothesis (Fig. 1C).  Fig. 2A. These results imply that the introduction of miR-663 notably reduced the chemosensitivity of MDA-MB-231 cells to ADM, CTX, and DTX. In addition, the inhibition of MDA-MB-231/ADM cell growth by the chemotherapeutic drugs was significantly increased by transfection with miR-663 ASO (Fig. 2B). To further investigate the function of miR-663, we also examined the toxicity of ADM, CTX, and DTX in BT-474 and MCF-7 cells, in which miR-663 was differentially expressed. BT-474 cells expressing an miR-663 mimic increased viability in response to treatment with ADM, CTX, or DTX compared with the control group (Fig. 2C), whereas MCF-7 cells transfected with miR-663 ASO exhibited significantly decreased viability in response to chemotherapeutic drugs (Fig. 2D). The cells were collected to assess the ROS levels by FACS and the apoptosis ratios by annexin V/PI staining and the TUNEL assay. The results demonstrated that miR-663 ASO induced cell apoptosis, accompanied by an increase in ROS level (Fig. 3,  A and B). We also detected the effects of the miR-663 mimic and of miR-663 ASO on apoptosis and the ROS level in BT-474 cells and MCF-7 cells, respectively, upon treatment with CTX or DTX. Transfection with an miR-663 mimic notably decreased the apoptotic ratio and ROS levels of BT-474 cells undergoing treatment with CTX or DTX (Fig. 4A) compared with the control group, whereas transfection with miR-663 ASO increased the apoptotic ratio and ROS levels of MCF-7 cells (Fig. 4B) undergoing treatment with chemotherapeutic drugs. We further detected a change in apoptosis-related proteins after treatment of MDA-MB-231 with the miR-663 mimic, using the RayBio human apoptosis antibody array. As shown in Fig. 3C, MDA-MB-231 cells transfected with the miR-663 mimic exhibited an obvious down-regulation of several pro-apoptosis proteins, including Bad, HTRA, IGFBP2, and p27, indicating that miR-663 suppresses the apoptotic response to chemotherapeutic chemicals in breast cancer cells.

miR-663 Regulates ROS and Apoptosis in Human Breast
miR-663 Regulates the Chemosensitivity of Human Breast Cancer Cells by Targeting HSPG2-Candidate miR-663 targets were predicted with TargetScan sequence analysis software, and HSPG2 was chosen for further examination (Fig. 5A). To further identify whether HSPG2 in breast cancer cells responded to miR-663 through direct interactions with its 3Ј-UTR, we cloned the WT 3Ј-UTR of the putative miR-663 target (Fig. 5B, WT 3Ј-UTR (Wt)) or mutated sequences (mut1 (Mu-1), mut2 (Mu-2), mut3 (Mu-3), mut1ϩmut2 (Mu-1ϩ2), mut1ϩmut3 (Mu-1ϩ3), mut2ϩmut3 (Mu-2ϩ3), mut1ϩmut2ϩ mut3 (Mu-1ϩ2ϩ3)) into a reporter plasmid downstream of the luciferase gene. These plasmids, together with miR-663 or miRcontrol and the Renilla luciferase vector (pRL-TK) (internal control), were transiently transfected into MDA-MB-231 cells. The Dual-Luciferase reporter assay system was used to detect luciferase expression 48 h after transfection. The results showed that the miR-663 mimic attenuated the fluorescence driven by the WT 3Ј-UTR by more than 2-fold compared with the negative control, whereas the mut1ϩmut2ϩmut3 3Ј-UTR mutant was not affected by miR-663 (Fig. 5B). To further confirm the potential role of miR-663 in the regulation of HSPG2, we evaluated the HSPG2 mRNA and protein expression levels in MDA-MB-231 and MDA-MB-231/ADM cells in which the miR-663 mimic was introduced or miR-663 ASO treatment had suppressed the endogenous miR-663 activity. As shown in Fig. 5, C and D, the up-regulation of endogenous miR-663 by the miR-663 mimic led to a notable decrease in the HSPG2 mRNA and protein levels, whereas miR-663 ASO transfection up-regulated the HSPG2 mRNA and protein levels. This inverse correlation between miR-663 and HSPG2 protein and mRNA levels provided sufficient evidence to support our conclusions. Collectively, these data demonstrated that HSPG2 was a direct target of miR-663 in breast cancer cells.
To elucidate the functional contributions of HSPG2 to the chemosensitivity of breast cancer cells, we transfected MDA-MB-231 cells with siRNA1 to -4 to target HSPG2. The siRNA1 and siRNA4 sequences silenced the HSPG2 gene significantly (Fig. 5E). We examined the viability and apoptosis of HSPG2 siRNA-transfected MDA-MB-231 cells in response to treatment with ADM. The CCK8 assay and apoptosis analysis showed that the chemoresistance of MDA-MB-231 cells increased upon HSPG2 knockdown with siRNA (Fig. 5, F and  G). Similar results were observed in BT-474 cells transfected with HSPG2 siRNA (Fig. 5H).
miR-663 Overexpression in Breast Cancer Tissue Correlates with Chemoresistance-To investigate whether miR-663 expression is related to chemoresistance in patients with breast cancer, we enrolled 39 breast cancer patients who underwent six or eight cycles of neoadjuvant chemotherapy and exhibited CR or PR, defined as chemosensitive, or PD, defined as chemoresistant, according to RECIST criteria. We employed in situ hybridization with a digoxigenin-labeled locked nucleic acid miR-663 probe to detect miR-663 expression in breast cancer tissues biopsied prior to neoadjuvant chemotherapy and analyzed the correlation between miR-663 level and clinical response to chemotherapy. A scrambled oligonucleotide was used to demonstrate the specificity of the hybridization. As shown in Fig. 6A and Table 2, miR-663 expression was up-reg-ulated in chemoresistant breast cancer tissues but reduced in chemosensitive tissues (p Ͻ 0.001). Because HSPG2 is the target of miR-663, we further investigated the expression of HSPG2 in breast cancer tissues using immunohistochemistry. In contrast to miR-663, HSPG2 expression was higher in the chemosensitive than in the chemoresistant breast tumor tissues (Fig. 6A). We also investigated the correlation between miR-663 expression and chemoresistance using real-time RT-PCR in 15 patients with breast cancer (Table 3). Similarly, we found that miR-663 expression was significantly up-regulated in chemoresistant tumor tissues (Fig. 6B, p Ͻ 0.001). In addition, miR-663 expression was significantly positively correlated to tumor size (r ϭ 0.425, p ϭ 0.003) and has a strong negative relationship with Ki67 (r ϭ 0.463, p ϭ 0.042) but no significant correlations between miR-663 expression and ER, PR, or Her2 level ( Table 2). The above results indicated that the overexpression of miR-663 in tumor tissues was associated with chemoresistance and the poor survival of some patients with breast cancer.
DNA Methylation Regulates the Expression of miR-663-We collected 10 colonies from methylation-filtered MDA-MB-231 and MDA-MB-231/ADM cells. The bacterial colonies were maintained as suspensions in 10% sterile glycerol at 4°C. The miR-663 clones were selected by colony PCR on randomly selected colonies from each sample. Bisulfite sequencing PCR was used to analyze the methylation status of the chemoresistant cell line and the parental cell line. MDA-MB-231/ADM contained fewer methylated CpG sites than did the parental cell line MDA-MB-231 (Fig. 7A). Furthermore, qRT-PCR showed that the miR-663 expression level in MDA-MB-231 cells was increased 3-4-fold by 5-Aza-dC treatment (Fig. 7B), indicating that DNA methylation may play a functional role in the expression of miR-663.

DISCUSSION
Individualized management for tumor patients is one of the most important therapeutic principles. The implementation of adjuvant therapy, hormonal therapy, and chemotherapy has had a major impact on the disease-free survival and overall survival of breast cancer patients (14). Unfortunately, many patients with breast cancer who are diagnosed and treated properly still suffer from recurrence and ultimately die of this disease (15). The essential factors that cause recurrence remain unknown. Researchers have long searched for prognosis-related markers. For many years, tumor size was the only criterion available with which to classify breast cancer stage. Later, a histological classification system and TNM staging information were developed. A more recent approach to classifying breast cancer subgroups is based on gene expression profiling. However, this subclassification has proven to be limited because it is unable to define subgroups with similar prognostic and therapeutic characteristics. Abnormal miRNA expression has been reported in many types of cancer, and much attention has been focused on understanding the roles of miRNAs in modulating cancer development (16). miRNAs were found to be useful biomarkers of tumorigenesis (17)(18)(19). miR-663, which is located at human chromosome 20q11.1, is associated with cellular senescence, immunity, and cancer.
This miRNA is increasingly thought to act as a tumor suppressor, and it was shown to be down-regulated in human gastric cancer cells and to induce growth arrest through mitotic catastrophe (10). miR-663 is also reported to be up-regulated in nasopharyngeal carcinoma cells and was determined to function as an oncogene in NPC by directly targeting p21 (12). However, the role of miR-663 in breast cancer, especially its correlation with chemosensitivity, has been unclear. In this paper, we identified miR-663 as the most pronouncedly up-regulated miRNA in MDA-MB-231/ADM cells compared with MDA-MB-231 cells. By modulating the miR-663 level in breast cancer cells, we revealed that miR-663 mediated the chemoresistance of breast cancer cells through the anti-apoptosis pathway.
These observations were extended to clinical samples of breast cancer, where we found a positive correlation between miR-663 expression level and chemoresistance. In addition, it was exhibited that miR-663 expression was positively significantly associated with tumor size but negatively related to Ki67. In our opinion, the possible explanation is cancer stem cell percentage. As we know, the "cancer stem cell" is very powerful in self-renewal, differentiation, and drug resistance, but always in the G 0 stage, during which it seldom proliferates. The asymmetric division is another key characteristic that involves stem cell division to new daughter stem cell and cells in different proliferation periods at the same time. Thus, the higher per-centage of "stem cell" is the lower ratio of proliferation index. In fact, high Ki67 expression only used to define a subset of chemosensitive tumors within ER-positive breast cancer. Moreover, several studies show that it is not powerful enough to successfully address interaction between Ki67 and treatment efficacy in ER-positive breast cancer and is not predictive for cyclophosphamide and methotrexate (20). Here, it is worth noting that miR-663 might be a biomarker of chemosensitivity that is independent of ER, PR, and Her2 expression.
In our study, we detected a change in apoptosis-related proteins after treatment of MDA-MB-231 with the miR-663 mimic, using the RayBio Human Apoptosis Antibody Array. It exhibited an obvious down-regulation of several pro-apoptosis proteins, including Bad, HTRA, IGFBP2, and p27, indicating that miR-663 suppresses the apoptotic response to chemotherapeutic chemicals in breast cancer cells. Then we predicted candidate miR-663 targets with TargetScan sequence analysis software and a luciferase reporter assay, and HSPG2 was chosen for further examination. Our findings from cell lines and clinical samples suggest that the overexpression of miR-663 induced chemoresistance in breast cancer cells by down-regulating HSPG2. miR-663 was overexpressed in chemoresistant breast cancer tissues but reduced in chemosensitive tissues, and on the contrary, HSPG2 expression was higher in chemosensitive than in chemoresistant breast tumor tissues. This inverse correlation between miR-663 and HSPG2 protein and mRNA levels provided sufficient evidence to indicate that HSPG2 was a target of miR-663 in breast cancer cells.
In addition, we delved into the mechanisms underlying the transcriptional regulation of miR-663. Common transcriptional regulation modes include DNA methylation (21), histone modification, and transcription factor binding (22). DNA methylation is one of the major epigenetic modification mechanisms that regulate gene expression, genomic imprinting, developmental timing, and tumorigenesis (23,24). The meth-FIGURE 6. miR-663 overexpression in breast cancer tissue correlates with chemoresistance. A, representative images of in situ hybridization for miR-663 and immunohistochemical staining for HSPG2 in breast cancer tissues. Scrambled probe and isotype IgG at the same concentration was used as a negative control for in situ hybridization and immunohistochemical staining, respectively. miR-663 was overexpressed in chemoresistant breast cancer tissues but reduced in chemosensitive tissues. Strong miR-663 staining, shown as navy blue grains widely distributed in the cytoplasm, was observed in the chemoresistant samples, whereas weakly positive staining scattered throughout the cytoplasm was observed in the chemosensitive samples. HSPG2 expression was higher in chemosensitive than in chemoresistant breast tumor tissues. B, miR-663 expression was reduced in chemosensitive breast cancer tissues but up-regulated in chemoresistant tissues, as shown by real-time RT-PCR with an miR-663-specific primer in samples from 15 patients with breast cancer. ylation of CpG dinucleotides often results in transcriptional inactivation, which is important for the normal development or differentiation of various tissues. In this study, we compared the methylation levels of the CpG islands immediately upstream of pre-miR-663 in MDA-MB-231 and MDA-MB-231/ADM cells and found that there were fewer methylated CpG sites in the putative miR-663 promoter in MDA-MB-231/ADM cells. Furthermore, 5-Aza-dC treatment notably increased the miR-663 expression level in MDA-MB-231 cells, indicating that miR-663 is silenced in MDA-MB-231 cells through promoter hypermethylation. These results indicated that chemotherapy reagents could result in the demethylation of the miR-663 promoter and the up-regulation of the miR-663 expression level in breast cancer cells. This increase in miR-663 level mediates chemoresistance in breast cancer cells.
In summary, we identified the function, target, and upstream regulatory mechanism of miR-663, both in cell lines and in clinical samples, providing evidence that the demethylation of the miR-663 promoter induces miR-663 expression and that miR-663 targets HSPG2 to mediate the chemotherapeutic resistance of breast cancer cells. We extended the current knowledge by highlighting the role of miR-663 in the chemoresistance of breast cancer for the first time. Although chemotherapy is the backbone of systemic treatment for most malignancies, its efficacy is hindered by the development of drug resistance. Therefore, targeting the mechanisms involved in the chemoresistance of breast cancer will improve treatment efficacy. It is logical to predict that miR-663 could be a potential biomarker and that the inhibition of miR-663 may be a candidate strategy for new breast cancer therapies.